Comparison of Fuzzy Time Series Prediction Models with Different FCM
In the fuzzy time series prediction model,the universe is usually divided by FCM clustering algorithm,or the traditional Euclidean distance is replaced by the similarity distance(DTW)calculated by sample points and clustering centers,that is,the DTW-FCM clustering algorithm is obtained to divide the universe.Among them,the FCM clustering algorithm based on DTW(DTW-FCM)introduces a negative exponential variable to obtain better robustness,so an improved FCM clustering algorithm based on DTW(DTW-MFCM)is designed.Finally,the prediction case of enrollment in Alabama University and the example of total carbon dioxide emissions in the United States are applied to this model,and the prediction results are compared with other fuzzy time series prediction models.The results show that the prediction accuracy and stability of this model are obviously improved,which is better than other classical models.
fuzzy time seriesrobustnessDTW-MFCM algorithmprediction model
NETL
NSTL
万方数据
